Headline

Backend & ML Engineer | Distributed Systems · LLM Integration · RAG Pipelines | UIDAI National Hackathon | Open to SWE & AI/ML Roles

About

Backend engineer completing B.Tech (Electrical Engineering, GEC Dahod — June 2026) who pivoted deep into distributed systems, ML pipelines, and LLM integration. I build things that run in production — not just notebooks. Projects include a hybrid-encrypted offline UPI payment mesh (Flask, RSA-OAEP, AES-GCM, optimistic locking), a vector database built from scratch with HNSW + RAG pipeline achieving sub-200ms ANN latency, a national-scale analytics engine for UIDAI processing 5M+ records across 807 districts, a multithreaded Deep Packet Inspection engine parsing Ethernet/IP/TCP/TLS across 77+ packets, an NLP-powered meeting intelligence tracker, and an LLM-powered Python-to-Java code migration tool — all with working code and live demos. Seeking entry-level Software Engineering or ML Engineer roles where I can contribute to scalable backend systems, AI/LLM products, or data infrastructure. Currently strengthening DSA and system design. Self-taught SWE background built through 6 production-grade projects. GitHub: github.com/T786-eng

Location

Bengaluru

AI stack

gemini

langchain

ollama

openai

Work experience

Data Analyst — National Hackathon at UIDAI Aadhaar Innovation Challenge 2026, Feb 2026 - Feb 2026 · 1 mo, Remote, Built Project DRAM (Dynamic Resource Allocation Model), a national-scale analytics engine processing 5M+ records across 12 data sources and 807 districts. Implemented Z-score anomaly detection and a custom UER metric to classify districts into strategic infrastructure zones. Presented live Streamlit dashboard to a national-level judging panel.

ML Engineer Intern at Edunet Foundation, Jan 2026 - Feb 2026 · 2 mos, Hybrid, Engineered a production-ready Random Forest classification pipeline on 12,000 records, achieving 91.2% accuracy and 0.87 F1-score. Designed end-to-end modular ML workflows with Pandas and Scikit-learn, eliminating data leakage and maximizing cross-environment reproducibility.

Machine Learning Intern at 3SKill, Dec 2025 - Jan 2026 · 2 mos, Gujarat, India, Architected a fraud detection backend on 20,000+ financial transactions. Applied SMOTE oversampling to lift recall from 61% to 84%, achieving 92% accuracy and 0.89 ROC-AUC. Reduced false-negative rate by 38% on highly imbalanced real-world data through precision-targeted feature engineering.

Education

Bachelor of Technology - BTech, Electrical Engineering, Government Engineering College, Dahod 018

St. Stephen Sr. Sec. School

Public links

github: https://github.com/T786-eng